EDA

Set Up

#--Install / load packages
rm(list = ls())
pacman::p_load(sf, here, tmap, osmdata, tidyverse, data.table, rio, tidyverse, flextable, mapview, units, spdep, deldir, sp, rgeoda, leaflet, viridis, crosstalk, leaflet.extras, plotly)

#--Import street-level crime data
crime <- rio::import(here::here("3_output", "crime_2024-05-09.csv")) |>
    dplyr::mutate(category = stringr::str_replace_all(category, "-", " ")) |>
    sf::st_as_sf(coords = c("location.longitude", "location.latitude"), crs = 4326, dim = "XY") 
    #from 2021-04 to 2024-03

#--Import Barnet shapefile
bnt_shp <- sf::st_read(here("1_data", "9_geo", "bnt_lad.json"), crs = 4326, quiet = TRUE) |>
  st_make_valid()

#--Get bounding box 
bb <- st_bbox(bnt_shp) 

#--Filter crime that intersects or is in within Barnet file
#crime_bnt <- crime[which(st_covers(bnt_shp, crime, sparse = FALSE)),]
crime_bnt <- crime[which(st_intersects(bnt_shp, crime, sparse = FALSE)),]

#--Amend date column
crime_bnt$date <- as.Date(paste0(crime_bnt$month, "-01"))

#--Create shared_data
shared_data <- SharedData$new(crime_bnt)

Spatio-Temporal Distribution of Crimes by Category

Map by Category

#--Assign colour palette
n_pal <- length(unique(crime_bnt$category))
crime_pal <- leaflet::colorFactor(turbo(n_pal), crime_bnt$category)

#--Create filters
month_slider <- crosstalk::filter_slider("date", "Date", shared_data, ~date, width = "100%")
category_checkbox <- crosstalk::filter_checkbox("category", "Category", shared_data, ~category)

#--Create map
m_eda <- leaflet(shared_data) |>
  leaflet::addProviderTiles("CartoDB.Positron")|>
  leaflet::addCircleMarkers(color = ~crime_pal(category), radius = 2) |>
  leaflet::addLegend("bottomright", pal = crime_pal, values = ~category, title = "Category")

#--Pull everything together
m1 <- crosstalk::bscols(
  widths = c(12, 3, 9),
  month_slider,
  category_checkbox,
  m_eda
)

m1

Date

Category

anti social behaviour

bicycle theft

burglary

criminal damage arson

drugs

other crime

other theft

possession of weapons

public order

robbery

shoplifting

theft from the person

vehicle crime

violent crime

#saveRDS(m1, here("1_data", "m1.RDS"))

Trend by Category

#--Create timeseries data by category
ct_crime <- crime_bnt |>
    st_drop_geometry() |>
    group_by(category, date) |>
    tally() |>
    arrange(desc(n))

#--Create shared_data
shared_data_ct <- SharedData$new(ct_crime)

#--Create filter
month_slider_ct <- crosstalk::filter_slider("date", "Date", shared_data_ct, ~date, step = 1, width = "100%")

#--Trend plot
pl_trend <- plotly::plot_ly(shared_data_ct, x = ~date, y = ~n, color = ~category, colors = viridis_pal(option = "H")(14)) |>
  plotly::add_lines() |>
  plotly::layout(xaxis = list(title = ""),
                 yaxis = list(title = "Number of Crimes\n"))

#--Pull everything together
p1 <- crosstalk::bscols(
  widths = c(12, 12),
  month_slider_ct,
  pl_trend
)

p1

Date

#saveRDS(p1, here("1_data", "pl_trend.RDS"))

Overall Frequency

In the last three years

#--Create count data by category
ct_crime2 <- crime_bnt |>
    st_drop_geometry() |>
    group_by(category) |>
    tally() |>
    arrange(desc(n)) 

#--Frequency plot
pl_freq <- plot_ly(ct_crime2, x = ~stats::reorder(category, n, decreasing = TRUE), y = ~n, color = ~category, colors = viridis_pal(option = "H")(14)) |>
  plotly::add_bars()|>
  plotly::layout(xaxis = list(title = ""),
                 yaxis = list(title = "Number of Crimes\n"))

pl_freq

#saveRDS(pl_freq, here("1_data", "pl_freq.RDS"))

In the last 12 months

#--Create count data by category in the last 12 months
ct_crime3 <- crime_bnt |>
    st_drop_geometry() |>
    filter(date >= ymd("2023-04-01")) |>
    group_by(category) |>
    tally() |>
    arrange(desc(n)) 

#--Frequency plot
pl_freq2 <- plot_ly(ct_crime3, x = ~stats::reorder(category, n, decreasing = TRUE), y = ~n, color = ~category, colors = viridis_pal(option = "H")(14)) |>
  plotly::add_bars()|>
  plotly::layout(xaxis = list(title = ""),
                 yaxis = list(title = "Number of Crimes\n"))

pl_freq2

#saveRDS(pl_freq2, here("1_data", "pl_freq2.RDS"))